This thread is locked.Only browsing is available.
Top Page > Browsing
runtestL Problem
Date: 2015/12/07 18:24
Name: Takeshi Mizuno

I have a question concerning the opnemx test results.

I compiled OpenMX 3.7 & patch 10 with IntelCompiler 13.1.3, OpenMPI 1.6.5 and FFTW 3.3.4.
The compiler options:

----------------------------------------------------

MKLROOT = /opt/intel/composer_xe_2013/mkl
FFTWROOT = /usr/local/OpenMX/FFTW-3.3.4
MPIROOT = /usr/local/openmpi-intel131
CC = $(MPIROOT)/bin/mpicc -openmp -O1 -axCORE-AVX2,AVX -I$(FFTWROOT)/include -I$(MKLROOT)/include
FC = $(MPIROOT)/bin/mpif90 -openmp -O1 -axCORE-AVX2,AVX -I$(MKLROOT)/include
LIB = -L$(MPIROOT)/lib -lmpi_f77 -lmpi_f90 \
-L$(FFTWROOT)/lib -lfftw3 -L$(MKLROOT)/lib/intel64 -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core \
-L/opt/intel/composer_xe_2013/lib/intel64 -lifcore -liomp5 -lpthread

----------------------------------------------------


When I ran OpenMX 3.7 using the -runtestL option, I got wrong result on "3 large_example/CG15c-Kry.dat".

The results of "3 large_example/CG15c-Kry.dat" is the following.


----------------------------------------------------

3 large_example/CG15c-Kry.dat Elapsed time(s)= 225.32 diff Utot= 0.000004980471 diff Force= 0.000173739387

----------------------------------------------------

This difference is out of range of "the difference is within last seven digits".



When I ran OpenMX 3.6 using the -runtestL option with the same environment,
I got the fine results of "3 large_example/CG15c-Kry.dat".

The resules of OpenMX 3.6 is following.

----------------------------------------------------

3 large_example/CG15c-Kry.dat Elapsed time(s)= 390.14 diff Utot= 0.000000001922 diff Force= 0.000000001661

----------------------------------------------------


What to do with these results?
Ex. just ignore? change OpenMX version? or ...?
ƒƒ“ƒe
Page: [1]